Probably the tree and surrounding vegetation were flowering abundantly, attracting many insects from the area. I would say thousands of insects on a large tree is not unfeasible at all, perhaps even tens of thousands. Whether they were all bees, and whether the bees were really all bumble bees, is not clear.
You're right that 25,000 bumble bees in one place does sound logistically difficult to arrange. We can back-of-the-envelope it...
(edit: increased foraging distance with data from reference, which slightly changes the conclusion)
It would be ~150 colonies within a several kilometer radius. If we assume mean foraging distance of 10km [1], thats (a rather beautiful) 314km^2 of source area. We should probably also assume that the vast majority of bees in the area are not visiting those two trees, as there are likely other good sources in the area too. If we assume 1-5% of the local bees might be visiting the trees, there would have to be a colony density of ~10-50 colonies per km^2. That is high, but not impossible. Previous measurements of bumble bee colony density have estimated in the region of 0.5 to 20 per km^2 [1].
So, one of the following is probable: the number of bumble bees is exaggerated in the story, or they weren't all bumble bees, or some of my estimates are wrong. I would arrogantly assert the first two are more likely :)
1. https://ir.library.oregonstate.edu/xmlui/handle/1957/35082